canonical: https://jentic.com/apis/cloudworldmodel.ai/cloudworldmodel

# Cloud World Model — Agent API

The Cloud World Model API simulates how a cloud deployment behaves so an agent can model cost and performance before provisioning anything real. It creates simulations from scenario templates, runs them synchronously or step by step, injects traffic and node failures, and returns per-resource cost breakdowns validated against provider benchmarks. On top of the simulation it offers chaos testing, multi-cloud strategy exploration, autoscaling threshold prediction, reinforcement-learning environments, and natural-language analysis of a run. Metered operations can be paid per call with x402 USDC micro-payments on Base or Solana, or accessed with a bearer token.

## For AI agents

Simulate cloud deployments to model cost and performance, run chaos tests, explore multi-cloud strategies, and train over reinforcement-learning environments without provisioning real infrastructure. Pay per call with x402 or use a bearer token.

## Scope

Does not provision real cloud resources, manage live production infrastructure, or send bills. Use for simulating, testing, and optimizing cloud deployments only.

## Capabilities

- Create cloud infrastructure simulations from scenario templates and run them synchronously or step by step
- Inject traffic and node failures to stress-test a simulated deployment
- Break down per-resource cost and validate it against provider benchmarks
- Submit infrastructure optimization jobs and generate autoscaling threshold recommendations
- Explore multi-cloud deployment strategies and stream results as they compute
- Create reinforcement-learning environments over a simulation and advance them by action
- Generate natural-language explanations, bottleneck analysis, and troubleshooting for a simulation

## Use cases

### Agent-Driven Infrastructure Modeling Through Jentic

An AI agent planning a cloud change can model it in the Cloud World Model through Jentic before touching real infrastructure. The agent creates a simulation from a scenario, runs it, reads the cost breakdown and bottleneck analysis, and only then recommends a provisioning step. Because simulations are addressed by id, the agent's reach is easy to bound to one run.

Example prompt: Create a simulation from a scenario template, run it statelessly, and read the per-resource cost breakdown to compare against the current setup

### Pre-Deployment Cost and Performance Modeling

Before committing a deployment, a team models it and validates the projected cost and performance against provider benchmarks. The simulation returns a per-resource cost breakdown and accuracy validation, and right-sizing hints suggest cheaper resource choices that still meet the workload.

Example prompt: Fetch a right-sizing hint for a simulation, apply it, and validate the updated cost and performance accuracy against benchmarks

### Chaos and Failure Testing in Simulation

Resilience testing runs against the model instead of production. A workflow injects targeted node failures or a chaos test, observes how the simulated deployment degrades and recovers, and schedules failure injections to rehearse incident response.

Example prompt: Run a chaos test against a simulation, then run recovery for the failed resource and confirm the deployment returns to a healthy state

### Reinforcement-Learning Environments Over Cloud Simulations

The simulation can be turned into a reinforcement-learning environment so a policy learns to operate cloud infrastructure. An agent resets the environment, reads the observation, takes an action step, and submits eval-episode jobs to measure how a policy performs over many episodes.

Example prompt: Create an RL environment over a simulation, reset it, read the observation, and run one action step to advance the episode

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | `/simulations` | Create a cloud simulation |
| POST | `/simulations/stateless` | Run a stateless synchronous simulation |
| GET | `/simulations/{simulationId}/cost-breakdown` | Fetch the per-resource cost breakdown |
| POST | `/chaos/run` | Run a chaos engineering test |
| POST | `/analysis/optimize` | Submit an infrastructure optimization job |
| POST | `/multi-cloud/explore` | Search multi-cloud deployment strategies |
| POST | `/rl/environments` | Create an RL training environment |
| GET | `/scenarios` | Search available simulation templates |

## Key resources

- **Simulations** — Create, run, step, and inspect cloud simulations, including cost breakdowns and accuracy validation
- **Chaos and Failures** — Run chaos tests, inject traffic and node failures, schedule failure injections, and recover resources
- **Optimize and Predict** — Infrastructure optimization jobs, autoscaling threshold generation, and traffic forecast validation
- **Multi-Cloud Strategy** — Explore multi-cloud deployment strategies and fetch full, partial, or streamed results
- **RL Environments** — Create reinforcement-learning environments, reset, step, batch-step, and run eval-episode jobs
- **AI Jobs and Analysis** — Submit asynchronous AI analysis jobs and generate explanations, bottleneck analysis, and troubleshooting

## AI readiness

This API is usable in Jentic One now. Its AI-readiness score against Jentic's framework shows where it stands today and where improvements would make it even easier for agents to use.

- **Score:** 61 / 100
- **Maturity:** AI-Aware
- **Dimensions:**
  - Foundational Compliance: 41 / 100
  - Developer Experience & Jentic Compatibility: 68 / 100
  - AI-Readiness & Agent Experience: 57 / 100
  - Agent Usability: 93 / 100
  - Security: 53 / 100
  - AI Discoverability: 88 / 100
- **View full report:** https://jentic.com/apis/cloudworldmodel.ai/cloudworldmodel/scorecard
- **How the score is calculated:** https://docs.jentic.com/reference/api-readiness-framework/overview/
- **More about the dimensions:** https://docs.jentic.com/reference/api-readiness-framework/specification/#dimensional-model-overview

### Score it yourself

Every API in the directory is allowlisted, so you can re-score it with no key required.

- **Score your own API:** https://jentic.com/scorecard.md
- **Scoring CLI agent skill:** https://github.com/jentic/jentic-api-scorecard/blob/main/skills/jentic-api-scorecard/SKILL.md

```sh
npx @jentic/api-scorecard-cli score <openapi-url>
```

## Why Jentic

- **Setup:** Wiring the Cloud World Model API by hand means handling bearer tokens or wallet session JWTs for authenticated calls and x402 USDC micro-payments on Base or Solana for metered ones, and tracking which operations carry a price. Through Jentic you install once, import Cloud World Model from the API Directory, store the token once, and your agent calls it.
- **Permission scoping:** Cloud World Model puts the simulation id in the URL path (the /simulations/ prefix), so a rule can pin your agent to one simulation and the operations you choose. You choose the operations it may call, so destructive ones like deleting a failure injection or cancelling an RL environment are not included unless you add them.
- **Credential handling:** Your Cloud World Model token is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as simulate a cloud deployment or run a chaos test, and Jentic returns the matching Cloud World Model operation with its input schema so the agent calls the right endpoint without reading the reference docs.

## Related APIs

- **Gremlin** — A chaos engineering platform that injects failure into real systems rather than a simulation.
- **Spot by NetApp** — Cloud cost and capacity optimization applied to running workloads.
- **DigitalOcean** — Provisions the Droplets and resources a simulation models once you decide to deploy.

## FAQ

### What authentication does the Cloud World Model API use?

The Cloud World Model API authenticates with a bearer token, which may be an API key or a wallet session JWT, per its OpenAPI spec. Metered operations can instead be paid per call with an x402 USDC micro-payment in the X-PAYMENT header on Base, or with a Solana payment signature, and some discovery and status operations are open. Through Jentic the token is stored encrypted by your own Jentic One instance and injected at call time.

### Can I run chaos tests with the Cloud World Model API?

Yes. You can run a single chaos engineering test or a batch of them in parallel against a simulation, and inject deterministic traffic or targeted node failures to see how the modeled deployment responds. Recovery operations then run recovery for a single failed resource.

### Is the Cloud World Model API free to use?

Most operations are metered: the spec marks them with x402 prices you pay per call in USDC, while a handful of discovery and status operations carry no per-call price. Fetch the current prices from the x402 config operation before running a metered call, and do not assume every read is free.

### How do I simulate a cloud deployment with the Cloud World Model API through Jentic?

Install Jentic One, then import Cloud World Model from the API Directory and store your token once. Your agent searches by intent such as create a cloud simulation, and Jentic returns the create-simulation operation with its input schema so the agent supplies the scenario and reads the simulation back. To run it on your own infrastructure, install Jentic One from its GitHub repo.

### Is there a Cloud World Model API MCP server?

You don't need an MCP server to give your agent the Cloud World Model API. Jentic connects it directly from the API Directory: import it, store your token once, and your agent calls the simulation, chaos, and optimization operations on demand. Operations are discovered when the agent needs them, so none sit in its context until a call is made.

### Can I limit what my agent is allowed to do with the Cloud World Model API?

Yes. Because the simulation id sits in the URL path, you can write a rule that pins your agent to one simulation and allows only the operations it needs, such as creating steps and reading cost breakdowns. Deleting failure injections or cancelling RL environments stays out of that set unless you add them, and every call is logged.